Researchers into moral education, and ethics educators too, often find themselves between a rock and a hard place. On the one hand, we wish to know what the child will do beyond the narrow range of communicative functions carried out in a classroom, and to do this, we employ purely hypothetical problems, that is, problems that from the child's point of view are constructed merely of language. On the other hand, we find that when we construct hypothetical problems out of language, the child's response is likewise merely constructed of language. Our solution is to turn this problem into a postulate: we take "mere language" as the object of ethical study and construe grammar as a chain of choices, many with moral implications. Therefore, for example, issuing imperatives implies a different ethical stance from creating interrogatives, and both are different again from sharing information with declaratives; so too, presenting an experience as a doing involves being answerable for the deed in a way that is different from presenting it as a happening. We offer a modest demonstration of how this approach could be used in research in language and education.